This is the first investigation of the effects of obesity, and of weight loss following BS, on DNA methylation in the rectal mucosa and in cfDNA. As summarised in Table 2, obesity was associated with altered methylation of LINE1 (an index of global DNA methylation (15)) and of a panel of genes linking obesity, inflammation and CRC risk including IL6, LEP, POMC and MAPK7. These alterations were reversed (normalised) following weight loss induced by BS (Table 2). For LINE1, associations with obesity and changes in response to BS observed in DNA from rectal mucosa were paralleled by differences in cfDNA. However, DNA methylation patterns in rectal mucosal DNA and in cfDNA were less consistent for the other genes in our panel.
Associations between obesity and DNA methylation
The lower methylation of LINE1 in DNA from rectal mucosal biopsies (CpG sites 2 and 4) and in cfDNA (CpG site 5) in Non-obese participants is contrary to what we expected since LINE1 hypomethylation is associated with higher risk of CRC(16). In a study of methylation of DNA from white blood cells in young (mean age 23 years), apparently healthy individuals (n = 156), Marques-Rocha et al.(17) found that indicators of adiposity, such as skinfold thickness and total body fat, were lower among individuals with higher LINE-1 methylation. There was no difference in LINE1 methylation in cfDNA from blood between obese and non-obese individuals in the BOCABS Study. This observation was similar to the findings from two studies that investigated LINE1 methylation in whole blood and in leukocytes. Nicoletti et al.(18) reported no difference in LINE1 methylation in blood between 36 obese and 9 healthy controls. A similar finding was reported by Duggan et al.(19) who investigated LINE1 methylation in leukocytes from 300 obese participants and 59 controls.
On the contrary, methylation of LEP in both DNA from rectal mucosal biopsies and in cfDNA was higher in the Non-Obese group compared with obese participants. This observation is supported by findings from House et al.(20) who observed a negative association between BMI and LEP methylation in blood from 73 obese patients. Similarly, Obermann-Borst et al.(21) reported that higher BMI was associated with lower methylation of LEP in blood from 120 children.
There was no significant difference in IL6 methylation in DNA from rectal mucosa samples between Pre-Surgery Obese and Non-Obese participants. However, IL6 methylation was higher in cfDNA from Non-Obese individuals. IL6 is a key regulator of inflammation and higher concentrations are associated with obesity and chronic inflammation(22). If methylation of IL6 is associated with transcriptional silencing(23), lower levels of IL6 methylation would be expected to lead to higher expression and higher circulating concentrations of IL6 in obese individuals. However, Carraro et al.(24) reported that higher IL6 methylation in blood was associated with higher BMI. The apparent conflict between these observations may relate to the specific CpG sites at which methylation was quantified since not all CpG sites within the same genomic region respond in the same way to environmental influences(25).
Methylation of POMC in DNA from rectal mucosal biopsies was lower in Non-Obese than in Obese participants. This finding is in line with those by Kuehnen et al.(26), who found that POMC methylation, quantified in 171 obese and 90 normal weight adolescents, was hypermethylated in obese individuals. Given evidence that methylation may inhibit POMC expression(27), one would anticipate that the lower methylation in non-obese individuals would result in higher POMC expression. POMC has anti-inflammatory properties and is associated with reduced risk of CRC(28). Thus, higher methylation in the rectum of obese participants, if associated with increased POMC expression, could be a potential mechanism for higher CRC risk in those living with obesity.
MAPK7 encodes a kinase that is part of the MAPK pathway that regulates cell proliferation, differentiation and apoptosis(29) and is a candidate oncogene(30). In the present study, MAPK7 methylation was lower in non-obese than in obese individuals, which is contrary to what is expected as MAPK7 hypomethylation has been observed in CRC(31). These apparently conflicting observations could be reconciled if hypomethylation of MAPK7 occurred later in the oncogenic process and so would not be detectable in macroscopically normal tissue, as investigated here.
Effects of bariatric surgery on DNA methylation:
Our previously systematic review showed that 12 out of 15 included studies reported significant changes in DNA methylation after BS in diverse tissues, including blood and adipose tissues (10). However, none of the studies investigated DNA methylation in rectal mucosal samples or in cfDNA from blood.
LEP methylation in rectal mucosal samples from BOCABS participants was significantly higher 6 months after BS and BS normalized methylation levels of LEP in the rectal samples, as demonstrated by the similarity in LEP methylation between post-surgery obese and non-obese individuals. It is important to note that the BMI of the obese group 6 months after BS was still within the obese range (mean = 32.3 kg/m2) in comparison with the won-obese group (mean = 25.3 kg/m2). These changes in LEP methylation in the rectal mucosa might contribute to the protective effects of weight loss in decreasing risk of CRC. Since LEP is overexpressed in CRC, higher LEP methylation leading to reduced LEP expression may lower risk of CRC(32). However, these favourable changes were not observed in cfDNA where methylation of LEP was lower after BS. This might be explained by differential effects of BS on metabolic and inflammatory outcomes in different body systems which can affect LEP methylation in cfDNA(33).
BS also normalized POMC methylation in rectal biopsies after 6 months in comparison with non-obese participants. If this lower POMC methylation is associated with increased expression, this would be expected to lower CRC risk since POMC has anti-inflammatory properties(28).
LINE1 methylation was lower in the rectal mucosa at 6 months after BS in obese patients with no significant difference between the post-surgery obese and non-obese groups, suggesting that LINE1 methylation was restored to non-obese levels. Similarly, methylation of LINE1 in cfDNA from blood was reduced after BS. These findings differ from the results of studies included in our systematic review and meta-analysis (10), where no effects of BS on LINE1 methylation in blood were observed. Martin Nunez et al.(34) and Nicoletti et al.(18) did not find any changes in LINE1 methylation in whole blood or buffy coat, respectively, 6 months after BS. This might be explained by the difference in methylation between sample types (DNA from blood which is largely from DNA within cells versus cfDNA from blood). From a wider perspective, the finding of lower LINE1 methylation in non-obese individuals and in the post-surgery group compared with the obese group pre-surgery is contrary to what might be expected since LINE1 hypomethylation has been linked to increased CRC risk(16). The reasons for this unexpected finding remain to be investigated.
There was no significant change in IL6 methylation after BS in DNA from rectal mucosal biopsies but methylation of IL6 in cfDNA increased after BS. In our published meta-analysis(10), the effects of BS on IL6 methylation differed between individual studies with reports of no effect(35), decreased methylation(18) or increased methylation(36) of IL6 in blood after BS leading to no overall effect in the meta-analysis. These heterogeneous findings may due to differences in participant characteristics, in the follow-up duration, and in the degree of weight change achieved after BS in each of the studies. In addition, the specific effects on BS on different tissues are likely contributing factors to the differential effects on IL6 gene methylation across studies.